Re: experts question.

Thanks for you response. I will run the following sql you suggested next time.

No errors in bdump folders. If it was a latch issue, does Oracle clean up the latch eventually?

"Steve Ashmore" <sashmore_at_neonramp.com> wrote in message news:<uqp1f5cqfobp39_at_corp.supernews.com>...
> You need to do more checking.
> Check v$session_wait to see what the active and hung sessions are waiting
> on.
> It may have been a latch.
>
> If it is a latch you will then need to drill further down to see which
> process is holding the latch and what it is up to. (v$latchholder).
>
> Either way more reasearch will need to be done.
> It can take some time but you will have more ammo for the next time.
>
> Good Luck!
>
> Stephen C. Ashmore
> Brainbench MVP for Oracle Administration
> http://www.brainbench.com
>
> Author of: 'So You Want to be an Oracle DBA?'
>
>
>
> "richard" <srichardchung_at_cs.com> wrote in message
> news:6a87bade.0210150452.6a4527b3_at_posting.google.com...
> > symptom:
> >
> > 1) no locks on v$lock.
> > 2) select v$session showed most of the session are active, even the
> > clients quit their client programs(some normal exit, some forced
> > exit). Users session always in v$session unless we killed the Unix
> > process. So,the # of session were piling up.
> > 3) Users can login, and select on most tables are ok, some selected
> > tables will hang their clients.
> > 4) didn't check v$transaction.used_ublk for rollback.
> > 5) CPU idle was 70% (12 CPUs). Only one or two Oracle processes are
> > active in Top session.
> >
> > We restarted the Oracle and the problems gone.
> >
> > Can anyone shed some light on what happened in Oracle? So we can
> > prevent it next time.
> >
> > Thanks in advanced.
Received on Wed Oct 16 2002 - 15:44:32 CDT